com.gargoylesoftware.htmlunit.WebRequest#setAdditionalHeader

Source File: HtmlEmbed.java From htmlunit with Apache License 2.0

6 votes

/**
 * Saves this content as the specified file.
 * @param file the file to save to
 * @throws IOException if an IO error occurs
 */
public void saveAs(final File file) throws IOException {
    final HtmlPage page = (HtmlPage) getPage();
    final WebClient webclient = page.getWebClient();

    final URL url = page.getFullyQualifiedUrl(getAttributeDirect(SRC_ATTRIBUTE));
    final WebRequest request = new WebRequest(url);
    request.setCharset(page.getCharset());
    request.setAdditionalHeader(HttpHeader.REFERER, page.getUrl().toExternalForm());
    final WebResponse webResponse = webclient.loadWebResponse(request);

    try (OutputStream fos = Files.newOutputStream(file.toPath());
            InputStream content =  webResponse.getContentAsStream()) {
        IOUtils.copy(content, fos);
    }
}

Source File: HtmlImageInput.java From htmlunit with Apache License 2.0

6 votes

/**
 * <p>Downloads the image contained by this image element.</p>
 * <p><span style="color:red">POTENTIAL PERFORMANCE KILLER - DOWNLOADS THE IMAGE - USE AT YOUR OWN RISK</span></p>
 * <p>If the image has not already been downloaded, this method triggers a download and caches the image.</p>
 *
 * @throws IOException if an error occurs while downloading the image
 */
private void downloadImageIfNeeded() throws IOException {
    if (!downloaded_) {
        // HTMLIMAGE_BLANK_SRC_AS_EMPTY
        final String src = getSrcAttribute();
        if (!"".equals(src)
                && !(hasFeature(HTMLIMAGE_BLANK_SRC_AS_EMPTY) && StringUtils.isBlank(src))) {
            final HtmlPage page = (HtmlPage) getPage();
            final WebClient webClient = page.getWebClient();

            final URL url = page.getFullyQualifiedUrl(src);
            final BrowserVersion browser = webClient.getBrowserVersion();
            final WebRequest request = new WebRequest(url, browser.getImgAcceptHeader(),
                                                            browser.getAcceptEncodingHeader());
            request.setCharset(page.getCharset());
            request.setAdditionalHeader(HttpHeader.REFERER, page.getUrl().toExternalForm());
            imageWebResponse_ = webClient.loadWebResponse(request);
        }

        downloaded_ = hasFeature(JS_IMAGE_COMPLETE_RETURNS_TRUE_FOR_NO_REQUEST)
                || (imageWebResponse_ != null && imageWebResponse_.getContentType().contains("image"));
    }
}

Source File: CsrfIT.java From krazo with Apache License 2.0

6 votes

/**
 * Checks that CSRF validation works if token sent as header instead of
 * form field.
 *
 * @throws Exception an error occurs or validation fails.
 */
@Test
public void testFormHeaderOk() throws Exception {
    HtmlPage page1 = webClient.getPage(webUrl + "resources/csrf");

    // Check response and CSRF header
    WebResponse res = page1.getWebResponse();
    assertEquals(Response.Status.OK.getStatusCode(), res.getStatusCode());
    assertNotNull(res.getResponseHeaderValue(CSRF_HEADER));

    WebRequest req = new WebRequest(new URL(webUrl + "resources/csrf"));
    req.setHttpMethod(HttpMethod.POST);
    req.setAdditionalHeader(CSRF_HEADER, res.getResponseHeaderValue(CSRF_HEADER));
    res = webClient.loadWebResponse(req);
    assertEquals(Response.Status.OK.getStatusCode(), res.getStatusCode());
}

Source File: Location.java From HtmlUnit-Android with Apache License 2.0

6 votes

/**
 * Reloads the current page, possibly forcing retrieval from the server even if
 * the browser cache contains the latest version of the document.
 * @param force if {@code true}, force reload from server; otherwise, may reload from cache
 * @throws IOException if there is a problem reloading the page
 * @see <a href="http://msdn.microsoft.com/en-us/library/ms536342.aspx">MSDN Documentation</a>
 */
@JsxFunction
public void reload(final boolean force) throws IOException {
    final HtmlPage htmlPage = (HtmlPage) getWindow(getStartingScope()).getWebWindow().getEnclosedPage();
    final WebRequest request = htmlPage.getWebResponse().getWebRequest();

    String referer = htmlPage.getUrl().toExternalForm();
    request.setAdditionalHeader(HttpHeader.REFERER, referer);

    referer = UrlUtils.getUrlWithNewQuery(htmlPage.getUrl(), null).toExternalForm();
    referer = StringUtils.stripEnd(referer, "/");
    request.setAdditionalHeader(HttpHeader.ORIGIN, referer);

    final WebWindow webWindow = window_.getWebWindow();
    webWindow.getWebClient().download(webWindow, "", request, true, false, "JS location.reload");
}

Source File: HtmlArea.java From HtmlUnit-Android with Apache License 2.0

5 votes

/**
 * {@inheritDoc}
 */
@Override
protected boolean doClickStateUpdate(final boolean shiftKey, final boolean ctrlKey) throws IOException {
    final HtmlPage enclosingPage = (HtmlPage) getPage();
    final WebClient webClient = enclosingPage.getWebClient();

    final String href = getHrefAttribute().trim();
    if (!href.isEmpty()) {
        final HtmlPage page = (HtmlPage) getPage();
        if (StringUtils.startsWithIgnoreCase(href, JavaScriptURLConnection.JAVASCRIPT_PREFIX)) {
            page.executeJavaScript(
                href, "javascript url", getStartLineNumber());
            return false;
        }
        final URL url;
        try {
            url = enclosingPage.getFullyQualifiedUrl(getHrefAttribute());
        }
        catch (final MalformedURLException e) {
            throw new IllegalStateException(
                    "Not a valid url: " + getHrefAttribute());
        }
        final WebRequest request = new WebRequest(url);
        request.setAdditionalHeader(HttpHeader.REFERER, page.getUrl().toExternalForm());
        final WebWindow webWindow = enclosingPage.getEnclosingWindow();
        webClient.getPage(
                webWindow,
                enclosingPage.getResolvedTarget(getTargetAttribute()),
                request);
    }
    return false;
}

Source File: ProducesIT.java From krazo with Apache License 2.0

5 votes

@Test
public void locale2() throws Exception {
    final WebRequest wrq = new WebRequest(new URL(webUrl + "resources/locale2"), ACCEPT_HEADER);
    wrq.setAdditionalHeader("Accept-Language", ACCEPT_LANGUAGE);
    final WebResponse wr = webClient.loadWebResponse(wrq);
    assertEquals(Response.Status.OK.getStatusCode(), wr.getStatusCode());
    assertEquals(MediaType.APPLICATION_XHTML_XML, wr.getContentType());
    assertEquals("en-GB", wr.getResponseHeaderValue("Content-Language"));
}

Source File: TrimouViewEngineTest.java From trimou with Apache License 2.0

5 votes

@Test
public void testController()
        throws FailingHttpStatusCodeException, IOException {
    WebClient webClient = new WebClient();
    WebRequest request = new WebRequest(
            new URL(contextPath, "resources/simple?user=mike"));
    request.setAdditionalHeader("Accept-Language", "cs");
    Page page = webClient.getPage(request);
    String[] parts = page.getWebResponse().getContentAsString().split(":");
    assertEquals(5, parts.length);
    assertTrue(parts[0].endsWith("/resources/simple"));
    assertEquals("mike", parts[1]);
    assertEquals("cs", parts[2]);
    assertEquals(parts[3], parts[4]);
}

Source File: ProducesIT.java From krazo with Apache License 2.0

5 votes

@Test
public void language1() throws Exception {
    final WebRequest wrq = new WebRequest(new URL(webUrl + "resources/language1"), ACCEPT_HEADER);
    wrq.setAdditionalHeader("Accept-Language", ACCEPT_LANGUAGE);
    final WebResponse wr = webClient.loadWebResponse(wrq);
    assertEquals(Response.Status.OK.getStatusCode(), wr.getStatusCode());
    assertEquals(MediaType.APPLICATION_XHTML_XML, wr.getContentType());
    assertEquals("es", wr.getResponseHeaderValue("Content-Language"));
}

Source File: HtmlLink.java From HtmlUnit-Android with Apache License 2.0

5 votes

/**
 * Returns the request which will allow us to retrieve the content referenced by the {@code href} attribute.
 * @return the request which will allow us to retrieve the content referenced by the {@code href} attribute
 * @throws MalformedURLException in case of problem resolving the URL
 */
public WebRequest getWebRequest() throws MalformedURLException {
    final HtmlPage page = (HtmlPage) getPage();
    final URL url = page.getFullyQualifiedUrl(getHrefAttribute());

    final WebRequest request = new WebRequest(url);

    request.setAdditionalHeader(HttpHeader.REFERER, page.getUrl().toExternalForm());

    final String accept = page.getWebClient().getBrowserVersion().getCssAcceptHeader();
    request.setAdditionalHeader(HttpHeader.ACCEPT, accept);

    return request;
}

Source File: ProducesIT.java From ozark with Apache License 2.0

5 votes

@Test
public void language1() throws Exception {
    final WebRequest wrq = new WebRequest(new URL(webUrl + "resources/language1"), ACCEPT_HEADER);
    wrq.setAdditionalHeader("Accept-Language", ACCEPT_LANGUAGE);
    final WebResponse wr = webClient.loadWebResponse(wrq);
    assertEquals(Response.Status.OK.getStatusCode(), wr.getStatusCode());
    assertEquals(MediaType.APPLICATION_XHTML_XML, wr.getContentType());
    assertEquals("es", wr.getResponseHeaderValue("Content-Language"));
}

Source File: HtmlEmbed.java From HtmlUnit-Android with Apache License 2.0

5 votes

/**
 * Saves this content as the specified file.
 * @param file the file to save to
 * @throws IOException if an IO error occurs
 */
public void saveAs(final File file) throws IOException {
    final HtmlPage page = (HtmlPage) getPage();
    final WebClient webclient = page.getWebClient();

    final URL url = page.getFullyQualifiedUrl(getAttributeDirect("src"));
    final WebRequest request = new WebRequest(url);
    request.setAdditionalHeader(HttpHeader.REFERER, page.getUrl().toExternalForm());
    final WebResponse webResponse = webclient.loadWebResponse(request);

    try (FileOutputStream fos = new FileOutputStream(file);
            InputStream content =  webResponse.getContentAsStream()) {
        IOUtils.copy(content, fos);
    }
}

Source File: Location.java From htmlunit with Apache License 2.0

5 votes

/**
 * Reloads the current page, possibly forcing retrieval from the server even if
 * the browser cache contains the latest version of the document.
 * @param force if {@code true}, force reload from server; otherwise, may reload from cache
 * @throws IOException if there is a problem reloading the page
 * @see <a href="http://msdn.microsoft.com/en-us/library/ms536342.aspx">MSDN Documentation</a>
 */
@JsxFunction
public void reload(final boolean force) throws IOException {
    final HtmlPage htmlPage = (HtmlPage) getWindow(getStartingScope()).getWebWindow().getEnclosedPage();
    final WebRequest request = htmlPage.getWebResponse().getWebRequest();

    if (getBrowserVersion().hasFeature(JS_LOCATION_RELOAD_REFERRER)) {
        final String referer = htmlPage.getUrl().toExternalForm();
        request.setAdditionalHeader(HttpHeader.REFERER, referer);
    }

    final WebWindow webWindow = window_.getWebWindow();
    webWindow.getWebClient().download(webWindow, "", request, true, false, "JS location.reload");
}

Source File: ProducesIT.java From ozark with Apache License 2.0

5 votes

@Test
public void locale1() throws Exception {
    final WebRequest wrq = new WebRequest(new URL(webUrl + "resources/locale1"), ACCEPT_HEADER);
    wrq.setAdditionalHeader("Accept-Language", ACCEPT_LANGUAGE);
    final WebResponse wr = webClient.loadWebResponse(wrq);
    assertEquals(Response.Status.OK.getStatusCode(), wr.getStatusCode());
    assertEquals(MediaType.APPLICATION_XHTML_XML, wr.getContentType());
    assertEquals("en-GB", wr.getResponseHeaderValue("Content-Language"));
}

Source File: Location.java From HtmlUnit-Android with Apache License 2.0

5 votes

/**
 * Sets the location URL to an entirely new value.
 * @param newLocation the new location URL
 * @throws IOException if loading the specified location fails
 * @see <a href="http://msdn.microsoft.com/en-us/library/ms533867.aspx">MSDN Documentation</a>
 */
@JsxSetter
public void setHref(final String newLocation) throws IOException {
    final HtmlPage page = (HtmlPage) getWindow(getStartingScope()).getWebWindow().getEnclosedPage();
    if (newLocation.startsWith(JavaScriptURLConnection.JAVASCRIPT_PREFIX)) {
        final String script = newLocation.substring(11);
        page.executeJavaScript(script, "new location value", 1);
        return;
    }
    try {
        URL url = page.getFullyQualifiedUrl(newLocation);
        // fix for empty url
        if (StringUtils.isEmpty(newLocation)) {
            final boolean dropFilename = page.getWebClient().getBrowserVersion().
                    hasFeature(ANCHOR_EMPTY_HREF_NO_FILENAME);
            if (dropFilename) {
                String path = url.getPath();
                path = path.substring(0, path.lastIndexOf('/') + 1);
                url = UrlUtils.getUrlWithNewPath(url, path);
                url = UrlUtils.getUrlWithNewRef(url, null);
            }
            else {
                url = UrlUtils.getUrlWithNewRef(url, null);
            }
        }

        final WebRequest request = new WebRequest(url);
        request.setAdditionalHeader(HttpHeader.REFERER, page.getUrl().toExternalForm());

        final WebWindow webWindow = window_.getWebWindow();
        webWindow.getWebClient().download(webWindow, "", request, true, false, "JS set location");
    }
    catch (final MalformedURLException e) {
        LOG.error("setHref('" + newLocation + "') got MalformedURLException", e);
        throw e;
    }
}

Source File: Downloader.java From MMDownloader with Apache License 2.0

5 votes

/**
 * HtmlUnit을 이용한 HTML 코드 파싱.
 *
 * @param eachArchiveAddress 실제 만화가 담긴 아카이브 주소
 * @return 성공 시 html 코드를 리턴
 */
private String getHtmlPageHtmlUnit(String eachArchiveAddress) throws Exception {
	/* 필수! 로그 메세지 출력 안함 -> HtmlUnit 이용시 Verbose한 로그들이 너무 많아서 다 끔 */
	java.util.logging.Logger.getLogger("com.gargoylesoftware").setLevel(Level.OFF);
	System.setProperty("org.apache.commons.logging.Log", "org.apache.commons.logging.impl.NoOpLog");

	print.info("일반 연결 시도중...\n");

	WebClient webClient = new WebClient();
	webClient.getOptions().setRedirectEnabled(true);

	WebRequest req = new WebRequest(new URL(eachArchiveAddress));
	req.setHttpMethod(HttpMethod.POST);
	req.setAdditionalHeader("User-Agent", UserAgent.getUserAgent());
	req.setAdditionalHeader("Accept-Encoding", "gzip"); //20171126 gzip 추가
	req.getRequestParameters().add(new NameValuePair("pass", PASSWORD)); //비밀번호 post 방식 전송

	HtmlPage page = webClient.getPage(req);

	//Html코드를 포함한 페이지 소스코드가 담길 스트링
	String pageSource = page.asXml();

	/** 여기도 페이지 파싱 실패 시 검증하는 코드 들어가야 됨 **/

	webClient.close();
	print.info("일반 연결 성공\n");
	return pageSource;
}

Source File: RequestLocaleSupportTest.java From trimou with Apache License 2.0

5 votes

@Test
public void testRequestLocaleSupport()
        throws FailingHttpStatusCodeException, MalformedURLException,
        IOException {
    WebClient webClient = new WebClient();
    WebRequest request = new WebRequest(new URL(contextPath, "test"));
    request.setAdditionalHeader("Accept-Language", "cs");
    TextPage page = webClient.getPage(request);
    assertEquals("cs:::fr", page.getContent());
}

Source File: Location.java From htmlunit with Apache License 2.0

5 votes

/**
 * Sets the location URL to an entirely new value.
 * @param newLocation the new location URL
 * @throws IOException if loading the specified location fails
 * @see <a href="http://msdn.microsoft.com/en-us/library/ms533867.aspx">MSDN Documentation</a>
 */
@JsxSetter
public void setHref(final String newLocation) throws IOException {
    final HtmlPage page = (HtmlPage) getWindow(getStartingScope()).getWebWindow().getEnclosedPage();
    if (newLocation.startsWith(JavaScriptURLConnection.JAVASCRIPT_PREFIX)) {
        final String script = newLocation.substring(11);
        page.executeJavaScript(script, "new location value", 1);
        return;
    }
    try {
        URL url = page.getFullyQualifiedUrl(newLocation);
        // fix for empty url
        if (StringUtils.isEmpty(newLocation)) {
            final boolean dropFilename = page.getWebClient().getBrowserVersion().
                    hasFeature(ANCHOR_EMPTY_HREF_NO_FILENAME);
            if (dropFilename) {
                String path = url.getPath();
                path = path.substring(0, path.lastIndexOf('/') + 1);
                url = UrlUtils.getUrlWithNewPath(url, path);
                url = UrlUtils.getUrlWithNewRef(url, null);
            }
            else {
                url = UrlUtils.getUrlWithNewRef(url, null);
            }
        }

        final WebRequest request = new WebRequest(url);
        request.setAdditionalHeader(HttpHeader.REFERER, page.getUrl().toExternalForm());

        final WebWindow webWindow = window_.getWebWindow();
        webWindow.getWebClient().download(webWindow, "", request, true, false, "JS set location");
    }
    catch (final MalformedURLException e) {
        if (LOG.isErrorEnabled()) {
            LOG.error("setHref('" + newLocation + "') got MalformedURLException", e);
        }
        throw e;
    }
}

Source File: HtmlForm.java From HtmlUnit-Android with Apache License 2.0

4 votes

/**
 * <span style="color:red">INTERNAL API - SUBJECT TO CHANGE AT ANY TIME - USE AT YOUR OWN RISK.</span><br>
 *
 * Gets the request for a submission of this form with the specified SubmittableElement.
 * @param submitElement the element that caused the submit to occur
 * @return the request
 */
public WebRequest getWebRequest(final SubmittableElement submitElement) {
    final HtmlPage htmlPage = (HtmlPage) getPage();
    final List<NameValuePair> parameters = getParameterListForSubmit(submitElement);
    final HttpMethod method;
    final String methodAttribute = getMethodAttribute();
    if ("post".equalsIgnoreCase(methodAttribute)) {
        method = HttpMethod.POST;
    }
    else {
        if (!"get".equalsIgnoreCase(methodAttribute) && StringUtils.isNotBlank(methodAttribute)) {
            notifyIncorrectness("Incorrect submit method >" + getMethodAttribute() + "<. Using >GET<.");
        }
        method = HttpMethod.GET;
    }

    final BrowserVersion browser = getPage().getWebClient().getBrowserVersion();
    String actionUrl = getActionAttribute();
    String anchor = null;
    String queryFromFields = "";
    if (HttpMethod.GET == method) {
        if (actionUrl.contains("#")) {
            anchor = StringUtils.substringAfter(actionUrl, "#");
        }
        final Charset enc = getPage().getCharset();
        queryFromFields =
            URLEncodedUtils.format(Arrays.asList(NameValuePair.toHttpClient(parameters)), enc);

        // action may already contain some query parameters: they have to be removed
        actionUrl = StringUtils.substringBefore(actionUrl, "#");
        actionUrl = StringUtils.substringBefore(actionUrl, "?");
        parameters.clear(); // parameters have been added to query
    }
    URL url;
    try {
        if (actionUrl.isEmpty()) {
            url = WebClient.expandUrl(htmlPage.getUrl(), actionUrl);
        }
        else {
            url = htmlPage.getFullyQualifiedUrl(actionUrl);
        }

        if (!queryFromFields.isEmpty()) {
            url = UrlUtils.getUrlWithNewQuery(url, queryFromFields);
        }

        if (HttpMethod.GET == method && browser.hasFeature(FORM_SUBMISSION_URL_WITHOUT_HASH)
                && WebClient.URL_ABOUT_BLANK != url) {
            url = UrlUtils.getUrlWithNewRef(url, null);
        }
        else if (HttpMethod.POST == method
                && browser.hasFeature(FORM_SUBMISSION_URL_WITHOUT_HASH)
                && WebClient.URL_ABOUT_BLANK != url
                && StringUtils.isEmpty(actionUrl)) {
            url = UrlUtils.getUrlWithNewRef(url, null);
        }
        else if (anchor != null
                && WebClient.URL_ABOUT_BLANK != url) {
            url = UrlUtils.getUrlWithNewRef(url, anchor);
        }
    }
    catch (final MalformedURLException e) {
        throw new IllegalArgumentException("Not a valid url: " + actionUrl);
    }

    final WebRequest request = new WebRequest(url, method);
    request.setAdditionalHeader(HttpHeader.ACCEPT, browser.getHtmlAcceptHeader());
    request.setAdditionalHeader(HttpHeader.ACCEPT_ENCODING, "gzip, deflate");
    request.setRequestParameters(parameters);
    if (HttpMethod.POST == method) {
        request.setEncodingType(FormEncodingType.getInstance(getEnctypeAttribute()));
    }
    request.setCharset(getSubmitCharset());

    String referer = htmlPage.getUrl().toExternalForm();
    request.setAdditionalHeader(HttpHeader.REFERER, referer);

    if (HttpMethod.POST == method
            && browser.hasFeature(FORM_SUBMISSION_HEADER_ORIGIN)) {
        referer = StringUtils.stripEnd(referer, "/");
        request.setAdditionalHeader(HttpHeader.ORIGIN, referer);
    }
    if (HttpMethod.POST == method
            && browser.hasFeature(FORM_SUBMISSION_HEADER_CACHE_CONTROL_MAX_AGE)) {
        request.setAdditionalHeader(HttpHeader.CACHE_CONTROL, "max-age=0");
    }
    if (browser.hasFeature(FORM_SUBMISSION_HEADER_CACHE_CONTROL_NO_CACHE)) {
        request.setAdditionalHeader(HttpHeader.CACHE_CONTROL, "no-cache");
    }

    return request;
}

Source File: HtmlUnitDownloder.java From gecco-htmlunit with MIT License

4 votes

public HttpResponse download(HttpRequest request, int timeout) throws DownloadException {
	try {
		URL url = new URL(request.getUrl());
		WebRequest webRequest = new WebRequest(url);
		webRequest.setHttpMethod(HttpMethod.GET);
		if(request instanceof HttpPostRequest) {//post
			HttpPostRequest post = (HttpPostRequest)request;
			webRequest.setHttpMethod(HttpMethod.POST);
			List<NameValuePair> requestParameters = new ArrayList<NameValuePair>();
			for(Map.Entry<String, Object> entry : post.getFields().entrySet()) {
				NameValuePair nvp = new NameValuePair(entry.getKey(), entry.getValue().toString());
				requestParameters.add(nvp);
			}
			webRequest.setRequestParameters(requestParameters);	
		}
		//header
		boolean isMobile = SpiderThreadLocal.get().getEngine().isMobile();
		webRequest.setAdditionalHeader("User-Agent", UserAgent.getUserAgent(isMobile));
		webRequest.setAdditionalHeaders(request.getHeaders());
		//proxy
		HttpHost proxy = Proxys.getProxy();
		if(proxy != null) {
			webRequest.setProxyHost(proxy.getHostName());
			webRequest.setProxyPort(proxy.getPort());
		}
		//timeout
		this.webClient.getOptions().setTimeout(timeout);
		//request,response
		webClient.getPage(webRequest);
		HtmlPage page = webClient.getPage(request.getUrl());
		HttpResponse resp = new HttpResponse();
		WebResponse webResponse = page.getWebResponse();
		int status = webResponse.getStatusCode();
		resp.setStatus(status);
		if(status == 302 || status == 301) {
			String redirectUrl = webResponse.getResponseHeaderValue("Location");
			resp.setContent(UrlUtils.relative2Absolute(request.getUrl(), redirectUrl));
		} else if(status == 200) {
			String content = page.asXml();
			resp.setContent(content);
			resp.setRaw(webResponse.getContentAsStream());
			String contentType = webResponse.getContentType();
			resp.setContentType(contentType);
			String charset = getCharset(request.getCharset(), contentType);
			resp.setCharset(charset);
		} else {
			throw new DownloadException("ERROR : " + status);
		}
		return resp;
	} catch(Exception ex) {
		throw new DownloadException(ex);
	}
}

Source File: HtmlUnitPageLoader.java From xxl-crawler with GNU General Public License v3.0

4 votes

@Override
public Document load(PageRequest pageRequest) {
    if (!UrlUtil.isUrl(pageRequest.getUrl())) {
        return null;
    }

    WebClient webClient = new WebClient();
    try {
        WebRequest webRequest = new WebRequest(new URL(pageRequest.getUrl()));

        // 请求设置
        webClient.getOptions().setUseInsecureSSL(true);
        webClient.getOptions().setJavaScriptEnabled(true);
        webClient.getOptions().setCssEnabled(false);
        webClient.getOptions().setThrowExceptionOnScriptError(false);
        webClient.getOptions().setThrowExceptionOnFailingStatusCode(false);
        webClient.getOptions().setDoNotTrackEnabled(false);
        webClient.getOptions().setUseInsecureSSL(!pageRequest.isValidateTLSCertificates());

        if (pageRequest.getParamMap() != null && !pageRequest.getParamMap().isEmpty()) {
            for (Map.Entry<String, String> paramItem : pageRequest.getParamMap().entrySet()) {
                webRequest.getRequestParameters().add(new NameValuePair(paramItem.getKey(), paramItem.getValue()));
            }
        }
        if (pageRequest.getCookieMap() != null && !pageRequest.getCookieMap().isEmpty()) {
            webClient.getCookieManager().setCookiesEnabled(true);
            for (Map.Entry<String, String> cookieItem : pageRequest.getCookieMap().entrySet()) {
                webClient.getCookieManager().addCookie(new Cookie("", cookieItem.getKey(), cookieItem.getValue()));
            }
        }
        if (pageRequest.getHeaderMap() != null && !pageRequest.getHeaderMap().isEmpty()) {
            webRequest.setAdditionalHeaders(pageRequest.getHeaderMap());
        }
        if (pageRequest.getUserAgent() != null) {
            webRequest.setAdditionalHeader("User-Agent", pageRequest.getUserAgent());
        }
        if (pageRequest.getReferrer() != null) {
            webRequest.setAdditionalHeader("Referer", pageRequest.getReferrer());
        }

        webClient.getOptions().setTimeout(pageRequest.getTimeoutMillis());
        webClient.setJavaScriptTimeout(pageRequest.getTimeoutMillis());
        webClient.waitForBackgroundJavaScript(pageRequest.getTimeoutMillis());

        // 代理
        if (pageRequest.getProxy() != null) {
            InetSocketAddress address = (InetSocketAddress) pageRequest.getProxy().address();
            boolean isSocks = pageRequest.getProxy().type() == Proxy.Type.SOCKS;
            webClient.getOptions().setProxyConfig(new ProxyConfig(address.getHostName(), address.getPort(), isSocks));
        }

        // 发出请求
        if (pageRequest.isIfPost()) {
            webRequest.setHttpMethod(HttpMethod.POST);
        } else {
            webRequest.setHttpMethod(HttpMethod.GET);
        }
        HtmlPage page = webClient.getPage(webRequest);

        String pageAsXml = page.asXml();
        if (pageAsXml != null) {
            Document html = Jsoup.parse(pageAsXml);
            return html;
        }
    } catch (IOException e) {
        logger.error(e.getMessage(), e);
    } finally {
        if (webClient != null) {
            webClient.close();
        }
    }
    return null;
}

Java Code Examples for com.gargoylesoftware.htmlunit.WebRequest#setAdditionalHeader()